Supervised morphogenesis: exploiting morphological flexibility of self-assembling multirobot systems through cooperation with aerial robots

Nithin MathewsAnders Lyhne ChristensenAlessandro StranieriAlexander ScheidlerMarco Dorigo


Abstract


Self-assembling robots have the potential to undergo autonomous morphological adaptation. However, due to the simplicity in their hardware makeup and their limited perspective of the environment, self-assembling robots are often not able to reach their potential and adapt their morphologies to task or environment without external cues or prior information. In this paper, we present supervised morphogenesis --- a control methodology that makes self-assembling robots truly flexible by enabling aerial robots to exploit their elevated position and better view of the environment to initiate and control (hence supervised) morphology formation on the ground. We present results of two case studies in which we assess the feasibility of the presented methodology using real robotic hardware. In the case studies, we implemented supervised morphogenesis using two different aerial platforms and up to six self-assembling autonomous robots. We furthermore quantify the benefits attainable for self-assembling robots through cooperation with aerial robots using simulation-based studies. The research presented in this paper is a significant step towards realizing the true potential of self-assembling robots by enabling autonomous morphological adaptation to a priori unknown tasks and environments.

Case study nr. 1





Case study nr. 2





Location-based group selection (LGS)


This is the implementation of the control methodology presented in this paper. The foot-bots do not have a priori knowledge about the environment or the target morphology to be formed. They are programmed to execute a phototaxis behavior until messages from the eye-bot are received. In case a gap wider than 5~cm is detected, the eye-bot starts to transmit messages to establish an STC link to a foot-bot that is approximately 40 cm away from the gap. All foot-bots remain static as long as messages are received from the eye-bot. Then, the STC link is extended by the eye-bot to include the foot-bot's neighbors required to form the target morphology. The number of neighbors to which the linked is extended to depends on the gap width. These foot-bots receive a SWARMORPH-script from the eye-bot and follow the instructions in the script to self-assemble into the target morphology of which the size depends on the width of the detected gap. Once the target morphology is formed, the foot-bots move towards the light to cross the gap.

Gap width 5cm Gap width 10cm
Gap width 15cm Gap width 25cm




Supervision based on random group (SRG)


This methodology allows us to isolate the performance benefits of selecting robots to form the target morphology based on their location in the environment. The foot-bots do not have a priori knowledge about the task or the target morphology. The foot-bots initially move towards the light until the eye-bot starts transmitting messages. In case a gap wider than 5~cm is detected, the eye-bot establishes an STC link to a random foot-bot, i.e., without considering its location in the environment with respect to the gap. The eye-bot repeats this process until the number of foot-bot with established STC links match the size of the target morphology. This group of randomly located foot-bots receive a SWARMORPH-script from the eye-bot containing the instructions necessary to form a target morphology that depends on the gap width. Once the morphology is formed, the foot-bots move towards the light to cross the gap.

Gap width 5cm Gap width 10cm
Gap width 15cm Gap width 25cm




Non-cooperative control (NCC)


This control methodology allows us to isolate the performance benefits of aerial supervision entirely. The foot-bots are pre-loaded with a SWARMORPH-script that they use to form a four foot-bot chain morphology when a gap (regardless of its actual width) is encountered. The foot-bots do not cooperate nor do they seek for supervision from the eye-bot. They initially move towards the light source until one of the foot-bots detects the gap using its ground sensors. The foot-bot warns neighboring foot-bots via messages sent over the mxRAB device and retreats approximately 40~cm from the gap. Subsequently, it invites neighboring foot-bots to connect to its rear. Other foot-bots stop executing the phototaxis behavior and volunteer to join the ongoing self-assembly process. Once the chain of four foot-bots is formed, the morphology moves towards the light source to cross the gap.

Gap width 5cm Gap width 10cm
Gap width 15cm Gap width 25cm